Millions of children with attention-deficit/hyperactivity disorder (ADHD) go undiagnosed for years, missing out on crucial early support. Now, an artificial intelligence tool could change that by spotting the hidden warning signs buried in routine medical records long before a formal diagnosis is made.
According to groundbreaking research published in Nature Mental Health, data scientists at Duke University School of Medicine have trained an AI model to analyse the electronic health records of more than 140,000 children. By reviewing a child’s medical history from birth through early childhood, the tool accurately estimates their risk of developing ADHD by age five and older.
Finding patterns in plain sight
The AI achieves this by recognising subtle combinations of developmental, behavioural, and clinical events that often appear years before a doctor officially recognises the condition. Crucially, the model maintained its high accuracy across all patient demographics, including sex, race, ethnicity, and insurance status.
“We have this incredibly rich source of information sitting in electronic health records,” explained Elliot Hill, the study’s lead author and a data scientist at Duke’s Department of Biostatistics & Bioinformatics. “The idea was to see whether patterns hidden in that data could help us predict which children might later be diagnosed with ADHD, well before that diagnosis usually happens.”
An early warning, not a doctor
The researchers stress that the AI is not designed to replace human judgment or make official clinical diagnoses. Instead, it acts as an early-warning system to flag at-risk children so primary care providers can refer them to specialists much earlier.
“This is not an AI doctor,” said senior author Dr Matthew Engelhard. “It’s a tool to help clinicians focus their time and resources, so kids who need help don’t fall through the cracks or wait years for answers.”
Study co-author Naomi Davis, PhD, an associate professor of psychiatry, noted that children with ADHD can struggle severely when their needs are not understood. By connecting families with evidence-based interventions earlier, the AI tool could help lay a much stronger foundation for a child’s future academic, social, and long-term health outcomes.
